Security audit
Security checks for vulnerabilities and agentic risk
This is a markdown-only Business English coaching skill with no code execution, credential access, persistence setup, or hidden data movement.
Install this if you want a focused workplace English coach. Be aware that it may activate broadly for business-English editing or practice requests, and users who prefer a specific dialect, tone, or non-native-like professional style should state that preference explicitly.
The phrase 'Trigger for any work-related English improvement request' is very broad and lacks clear boundaries for when the skill should or should not activate. It overlaps with many ordinary workplace requests and does not provide exclusion conditions or negative examples to limit accidental invocation.
The description says it 'drills the exact language patterns used by native English speakers at work,' which prescribes a specific linguistic norm rather than offering style options. This can constitute a language-policy issue because it frames one variety of English as the default target without documenting user choice or opt-in.
The README says to use the skill when someone wants to 'sound more professional in English at work' or improve workplace communication, then concludes with 'Trigger for any work-related English improvement request.' This describes a wide intent class rather than a precise activation rule, increasing the chance of collisions with common editing or communication requests.
The core scope repeats that the skill teaches 'the exact language patterns used by native English speakers at work.' Because this is presented as the default behavior, it imposes a specific language standard without indicating that users may prefer other valid professional English styles.
The skill description and trigger guidance are broad enough to activate on many generic work-related English requests, which can cause the agent to invoke this skill outside the user's precise intent. Over-broad invocation is risky because it can overshadow more appropriate skills, leading to incorrect handling of user requests or unintended exposure of user content to this skill's prompt logic.
No suspicious patterns detected.
